DESCRIPTION (Applicant's Abstract): In this competitive continuation application, the primary goal is to evaluate the antidepressant effects of testosterone compared to fluoxetine in a double-blind randomized placebo controlled clinical trial for HIV-positive patients with major depression as well as clinical symptoms of hypogonadism. In the first 2 years of the ongoing testosterone grant, the investigators treated HIV-positive men with clinical symptoms of hypogonadism in a 12-week open trial followed by a 6-week double-blind placebo controlled discontinuation trial with improved libido as the major outcome criterion. In the discontinuation phase, 79 percent (26/33) randomized to testosterone maintained response, compared to 15 percent (5/34) for placebo. Twenty-seven men in this sample had major depression. Of these, 21 (78 percent) had significantly improved mood after 8 weeks of open treatment. In the ongoing NIMH-supported placebo controlled trial of fluoxetine, 36 patients with major depression have been treated with fluoxetine, and 22 with placebo in a double-blind randomized trial. Response rates to fluoxetine and placebo respectively have been 75 percent and 45 percent. The results, taken together, suggest the value of a study designed to directly compare antidepressant effects of testosterone, fluoxetine, and placebo. The design will be an 8-week double blind 3-arm study of testosterone injections (and placebo capsules), fluoxetine (and placebo injections), and placebo (placebo injections and placebo capsules). The investigators propose a 2:1:1 randomization with 140 patients (70:35:35) completing the 8-week trial. The testosterone arm is the largest since they seek to determine whether an antidepressant effect is secondary to or independent of androgenic (increased libido) or anabolic (increased weight and/or muscle mass) effects. After this trial, treatment will be continued for another 4 months to assess longer-term patient acceptance, side effects, and safety, using HIV RNA viral load to gauge immune effects. The intent is to identify a practical, inexpensive, useful treatment for community settings.